NPR logo

Need Stock Tips? Read Your Tweets

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
Need Stock Tips? Read Your Tweets

Need Stock Tips? Read Your Tweets

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

GUY RAZ, host:

Now, if you don't have access to the halls of power but you still want some inside stock information, you might find it on Twitter. Johan Bollen is a professor at Indiana University where he studies the links between those brief messages sent out on Twitter every day and various social indicators, including the direction of the Dow Jones Industrial Average.

And he made the discovery after asking his graduate students to collect millions of tweets and analyze them not for the financial information they contained but...

Professor JOHAN BOLLEN (School of Informatics, Indiana University): Specifically the emotional content, the type of words that people naturally use to express their emotional state.

RAZ: So what kind of words are you looking for in tweets?

Prof. BOLLEN: Well, we've got a wide range of it. And the words that we're looking for has been derived from an existing psychological mood test. And so that allows us to look at about - at thousands of mood words, like for example: I'm very sad, I'm angry, I'm at ease.

RAZ: So if most people, based on the Twitter mood analysis, are happy, there's a good chance the stock market will go up that day?

Prof. BOLLEN: Well, that's not what we were expecting. We were expecting that if the market goes up, then afterwards, people will be happy about it. If the market goes down, then afterwards, people will be sad about it.

RAZ: Right.

Prof. BOLLEN: And so, the student walks into my office - and I still remember this vividly - and shows me these two curves, and that was Huina Mao, she's a co-author of one of our recent papers. And she said, well, I sort of plotted these two curves. One is the Dow Jones, the other one is one of our mood signals and they overlapped.

And I said, that's great, but that's entirely to be expected, because as I said, you know, the market does well, people happy; market does poorly, people sad. And she said, well, it's not happy-sad.

And that made me think for a second. And I said, well, why - what is it? And she said, it's calm versus anxious. It's a different dimension of mood. And that kind of surprised me.

And then she said, and there's one more thing you should know. I have to shift the mood curve forward in time by three to four days to have them match up.

And that told me - at that point, I realized, well, that means that the mood state is actually predicting the Dow Jones. And that's when we both, you know, sank into our chairs a little bit, and that's when I told her that this is big. This is astonishing.

RAZ: So how accurate is this?

Prof. BOLLEN: We found that three or four days out, we have an 86.7 percent accuracy in predicting whether the Dow Jones is going to go up and down.

RAZ: Wow.

Prof. BOLLEN: We tested this for a particular period in the fall of 2008 for two, three months.

RAZ: Which was a dark period on Wall Street, obviously.

Prof. BOLLEN: Right. Also a very volatile period. But we decided to test it specifically for that period because if it works for that period, then we figured it would work for later periods as well.

RAZ: Do we know why a general mood would affect the stock market rather than, say, basically economic fundamentals?

Prof. BOLLEN: Yeah. Our research doesn't actually say anything about that. But my sort of best conjecture is that these user communities have become so big that a lot of investors are actually involved on Twitter. And somehow they may pick up on the national and international zeitgeist, and then unconsciously, perhaps, that works itself into their investment strategies.

RAZ: So Johan, will you guys make day-to-day research publicly available? In other words, I mean, presumably, investors would want to know what you're finding out, right, so then they could choose to invest that day or choose to pull their money out that day?

Prof. BOLLEN: Yeah. I guess that could be done. Of course, you know, we're scientists, so we're not interested in making a quick buck on the Dow Jones Industrial Average. We're interested in what this phenomenon tells us about the social and economic relations between these online communication environments and real-life and real-world economic outcomes.

RAZ: Can you just give us a little tip? Where is the market going to go in the next three days?

(Soundbite of laughter)

Prof. BOLLEN: I'm sorry, that's just not going to happen.

RAZ: That's Johan Bollen. He's an associate professor in the School of Informatics and Computing at Indiana University in Bloomington. His research looks into the correlation between tweets on Twitter and the value of the Dow Jones Industrial Average.

Johan Bollen, thank you so much.

Mr. BOLLEN: My pleasure. It was great to be on the show.

Copyright © 2010 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.